Open Access
Add to BibSonomy
Abstract
We report a directly modulated laser diode system capable of generating picosecond-pulse bursts and nanosecond pulses simultaneously. A generated pulse shape can be arbitrary controlled with a temporal resolution of 1 ns and wavelength of 1064 nm. A two-stage Nd:YVO4 amplifier boosts the pulse energy to hundreds of microjoules to process JIS 304 stainless steel. Characterization of processed holes irradiated by different pulse durations and shapes reveals that the ablation efficiencies with the nanosecond pulses are two times higher than those with the picosecond-pulse bursts. A clear hole with a taper angle of 1.5° is realized by the picosecond-pulse bursts with a 10-ns pulse interval. The combination of pulses with large differences in timescales offers an efficient production line with a single laser system.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
The market for material processing using high-power lasers has grown in recent years. A proposed method of optimizing such processing uses a database of processing results in a computer simulation [1]. Laser parameters such as wavelength, pulse duration, optical power, and pulse energy greatly influence the quality of a laser-irradiated material surface. Many lasers with different parameters are used in manufacturing processes. It is possible to reduce the processing time by the combination of lasers with different parameters. Laser induced breakdown spectroscopy uses different pulse durations and shapes to improve the sensitivity [2]. Therefore, flexibility of laser parameters is an important factor for laser-related applications such as material processing, microscopy, and scientific research. Above all, the duration and shape of a laser pulse greatly affect laser–matter interaction.
Over the past few years, several approaches to arbitrary pulse shaping have been reported. Pulse shaping for a broad spectrum of ultrashort pulses was achieved via spectral filtering with a spatial light modulator to modify the intensity and phase of dispersed light [3,4]. This improves the signal-to-noise ratio of an optical microscope [5]. For pulses having a narrow spectrum, several other approaches have been proposed and demonstrated. The pump energy distribution emitted from a flash lamp was controlled by dividing a driving pulse into several sectors [6]. This enabled generation of different output pulse shapes with sub-millisecond temporal resolution. However, it is difficult to control a pulse shape precisely because it involves an energy transfer from an excited state to a lasing state. A pulse was shaped with a temporal resolution of several hundred picoseconds by combining a continuous-wave (CW) laser source, an electro-optic modulator, and an acousto-optic modulator [6–10]. However, a controllable range of this approach is limited to a few hundred nanoseconds. In addition, a residual CW component could result in a pedestal when combined with high-gain laser amplifier systems. A more direct and simple approach is direct modulation of a laser diode (LD), which achieves a temporal resolution of 4 ns [11]. This approach produces a wide range of pulse shapes from nanosecond to microsecond. In addition, the gain switch operation of LDs can generate a pulse of several tens of picoseconds [12,13]. Chen et al. demonstrated sub-nanosecond pulse bursts in the form of an arbitrary packet shape [14]. Picosecond-pulse bursts have the potential to improve material ablation rates [15]. Advanced material processing requires precise control of the laser input to a material. Therefore, the laser pulse shape needs to be more flexible to generate, for example, a combination of picosecond and nanosecond pulses with arbitrary shapes. However, no report has been found on such a flexible laser for practical applications.
In this paper, we demonstrate, for the first time to our knowledge, a temporally programmable laser system using a LD with direct current modulation that produces picosecond and nanosecond pulses simultaneously. A 1064-nm nanosecond pulse shape can be arbitrary controlled with 1-ns temporal resolution. The generated laser pulses are amplified by a two-stage ytterbium (Yb)-doped single-mode fiber (YDF) amplifier and a two-stage neodymium (Nd):YVO4 amplifier. We also report, for the first time to our knowledge, simultaneous material processing by picosecond-pulse bursts and nanosecond pulses generated by the laser system.
2. Experimental setup
A schematic of the experimental setup is shown in Fig. 1. A distributed feedback grating (DFB) LD operating at a wavelength of 1064 nm was used to generate seed pulses. The DFB-LD was driven by a current driver with a 4-GHz bandwidth connected to a digital-to-analog converter (DAC) with a 1-GHz bandwidth. The current driver generated arbitrary drive current pulses with a temporal resolution of 1 ns with 12-bit amplitude resolution. The temporal duration was programmable and could be varied up to a maximum of 262.144 µs (i.e., 262,144 programmable points). The maximum drive current was 1.0 A. A desired pulse shape was fed to the current driver by a personal computer via an interface board. A polarization-maintained (PM) single-mode fiber (SMF) with a mode field diameter of 6.6 µm at 980 nm was connected to the DFB-LD. The PM-SMF was connected to an optical isolator that was used to prevent back reflections from damaging the DFB-LD. This was followed by an amplifier consisting of a YDF with a mode field diameter of 6.0 µm at 1060 nm, a wavelength division multiplexer (WDM), an optical isolator (ISO), a bandpass filter (BPF), and the PM-SMF. The two stages of the YDF amplifier were used to increase the pulse energy to several microjoules.
Fig. 1. Schematic of the temporally programmable laser system. DFB-LD: distributed feedback grating laser diode; FBG-LD: fiber-Bragg-grating-stabilized LD; ISO: isolator; WDM: wavelength division multiplexer; YDF: ytterbium-doped fiber; BPF: bandpass filter; HWP: half-wave plate; PBS: polarizing beam splitter cube; AL: aspheric lens; L1–2: lenses; L3–4: collimation lenses; L5: focusing lens; FCLD: fiber-coupled LD; MS: mechanical shutter; PD1–2: photodiodes.
After the two-stage YDF amplifier, the end of the PM-SMF had a standard FC/APC connector connected to an aspheric lens (AL) as a collimator. The pump source for the two-stage YDF amplifier was a fiber-Bragg-grating-stabilized LD (FBG-LD) generating an output power of 450 mW at a wavelength of 976 nm. A 50-to-50 coupler was used to distribute the pump energy to each YDF amplifier. The net gain of the two-stage YDF was measured to be 42.6 dB when an input pulse had a Gaussian temporal profile with a full width at half maximum (FWHM) of 20.7 ns. The laser pulse profile was measured using a high-speed InGaAs photodetector (NewFocus, Model 1024) that was fiber-coupled with the PM-SMF. The bandwidth of the photodetector was 29 GHz. The measured electric signal was displayed on a 20-GHz oscilloscope (LeCroy, WaveMaster SDA-820Zi-A) with a rise time (10–90%) of 22 ps. Figure 2 shows the different shapes of laser pulses generated from the DFB-LD when a peak current of 500 mA was applied. The central wavelength in vacuum was adjusted to be 1064.4 nm at an operating temperature of 18.5°C. The pulse repetition frequency was set to 300 Hz. For generating nanosecond pulses, the bias current needs to be applied to induce carrier recombination as quickly as possible when the drive current is applied. Because the DFB-LD used in the experiment had a threshold current of 17.4 mA, the appropriate bias current was set to 15.7 mA.
Fig. 2. Temporal characteristics of the pulse shapes generated from the directly modulated DFB-LD. The applied peak current was 500 mA. (a) 4.9-ns pulse. (b) 20.5-ns pulse. (c) Gain switching operation generating 53-ps pulse. The inset shows a gain switching pulse with a pedestal due to too much bias current. (d) Combination of picosecond and nanosecond pulses.
Figures 2(a) and (b) show the generated nanosecond pulses with durations of 4.9 ns and 20.5 ns (FWHM), respectively. The corresponding pulse energies were measured with a Si photodiode to be 0.15 nJ and 0.63 nJ, respectively. Meanwhile, precise control of the bias current is required for generating a single picosecond pulse. If the bias current is high (Ib = 1.0 mA), a pedestal appears as shown in the inset of Fig. 2(c). A single picosecond pulse was obtained with a proper bias current setting (Ib = 0.6 mA) (Fig. 2(c)). The pulse duration was measured to be 53 ps (FWHM). Although the pulse duration generated by gain switching cannot be adjusted, the amplitude can be independently controlled.
Because the pulse energy was less than the detection limit of the detector, we estimated the pulse energy to be 10 pJ by indirectly measuring a nanosecond pulse energy with and without the picosecond pulse. The highly flexible pulse-shaping capability makes it possible to combine picosecond and nanosecond pulses as shown in Fig. 2(d). The bias current was set to 0.6 mA. The measured pulse energy was 0.31 nJ. We attribute the sharp rise at the beginning of the nanosecond pulse to relaxation oscillation of the DFB-LD.
Flexible pulse shaping allows one to control the energy applied to a material with different temporal characteristics with arbitrary timing. Sophisticated material processing could benefit from this capability. After the generated laser pulses were amplified by the two-stage YDF amplifier, the pulse energies were amplified to 0.5 µJ, 1.8 µJ, and 6.0 nJ, corresponding to the input pulse shapes in Fig. 2(a)–(c). We observed no significant change in pulse shape because of gain saturation in the YDF amplifiers.
The short- and long-term energy stability of the directly modulated LD was measured. The short-term stabilities for 20-ns gaussian and picosecond pulses were 0.62% rms and 4.07% rms. The reduced stability of the gain switching is attributed to mode competition around the threshold [16]. The long-term stability is shown in Fig. 3 for a 20-ns Gaussian temporal profile. Long-term stability was achieved with an energy fluctuation of 0.87% rms for over 2500 hours.
The output from the YDF amplifier was fed to a two-stage LD-pumped Nd:YVO4 amplifier. A half-wave plate (HWP) and a polarizing beam splitter (PBS) cube were used to control the input pulse energy to the Nd:YVO4 amplifier. Two 808-nm fiber-coupled LDs (FCLDs) were used as pump sources for the Nd:YVO4, providing a maximum pump power of 50 W. Spherical lenses (L1 and L2) were used for mode matching to the pump beams. The dimensions of the Nd:YVO4 crystal were 3.0 mm (width) by 3.0 mm (height) by 5.0 mm (length). The Nd doping concentration was 1.0 atomic percent. The front facet of the Nd:YVO4 crystal had an anti-reflection coating at a laser wavelength of 1064 nm, and both facets had an anti-reflection coating at a pumping wavelength of 808 nm. The typical gains of the first and second Nd:YVO4 amplifiers were measured to be 30 and 4, respectively, for input pulse energies of 4 µJ and 100 µJ. The amplified output was collimated by a pair of spherical lenses (L3-4).
3. Material processing with different pulse durations and shapes
Several pulse shapes after amplification were characterized as shown in Fig. 4. The insets of Fig. 4 show the corresponding shapes of the pulses input to the Nd:YVO4 amplifier, which were precisely controlled by the programmable current driver taking into account gain saturation. Figure 4(a) shows a standard square pulse with a duration of 100 ns and pulse energy of 185.0 µJ. Figures 4(b) and (c) incorporate gain switching into pulse bursts output. The pulse interval of Fig. 4(b) was set to 5 ns, which corresponded to a repetition frequency of 200 MHz. The pulse burst energy was measured to be 49.0 µJ (equivalent to 4.9 µJ/pulse). Meanwhile, the pulse interval of Fig. 4(c) was gradually changed from 3 ns to 19 ns with 2-ns increments, generating a burst pulse energy of 48.6 µJ. The maximum pulse burst repetition frequency was 500 MHz because separable pulses need at least a 2-ns pulse interval. Figure 4(d) shows the combination of picosecond-pulse bursts and a nanosecond pulse with a duration of 71.2 ns. The separation between the first picosecond pulse and the peak of the nanosecond pulse was 293 ns. The pulse energies of the picosecond-pulse bursts and the nanosecond pulse were 15.5 µJ and 238.5 µJ, respectively. The beam diameter was measured to be 8.0 mm (1/e2 intensity), which can be adjusted using the appropriate collimator lens L3. The measured M2 value was 1.4. The pointing stability was also measured to be within 20 µrad. The stability of the pulse energy with a 20-ns Gaussian profile was 0.95% rms for over 100 hours.
Fig. 4. Temporal characteristics of the output pulse generated from the Nd:YVO4 amplifiers. The insets show the corresponding input pulse shape. (a) 100-ns square pulse. (b) Picosecond-pulse bursts with 10 pulses. The pulse interval was kept constant at 5 ns. (c) Picosecond-pulse bursts with gradually changing pulse intervals. (d) Combination of picosecond-pulse bursts and nanosecond pulses.
The flexibility of the temporal profile of a laser pulse offers important advantages for material processing. To demonstrate this, material processing experiments were carried out using the developed laser. Austenitic stainless steel (JIS SUS304) was used as a sample material. The thickness of the sample was 50 µm. The sample was placed on a motorized two-axis translation stage. Linearly polarized laser pulses with different temporal profiles were focused onto the sample, which was drilled using a plano-convex focusing lens (L5) with a focal length of 40 mm (f/5). The focused spot diameter was calculated to be 10 µm. The pulse energy was adjusted to be 40 µJ for each temporal pulse shape. The laser irradiation number was kept constant at 900 shots by a mechanical shutter (MS). No assist gas was used in this experiment.
The quality of the drilled holes was inspected through scanning electron microscope (SEM) images. At least six holes were processed for each parameter to confirm reproducibility. Figure 5 shows SEM images of the hole entrances, and the insets show the corresponding hole exits. The temporal pulse profiles for Fig. 5(a)–(h) were a 100-ns rectangular pulse (Fig. 4(a)), 20-ns Gaussian pulse, combination of picosecond-pulse bursts and a 71.2-ns pulse (Fig. 4(d)), picosecond-pulse burst of 10 pulses with a 2-ns interval, picosecond-pulse burst of 10 pulses with a 5-ns interval (Fig. 4(b)), picosecond-pulse burst of 10 pulses with a 10-ns interval, picosecond-pulse burst of 10 pulses with a 50-ns interval, and a single picosecond pulse with a 3.3-ms interval, respectively. The ablated surface had different results depending on the temporal pulse shape.
Fig. 5. SEM images of the SUS304 processed using different patterns of laser pulses. (a) 100-ns rectangular pulse. (b) 20-ns Gaussian pulse. (c) Burst of five picosecond pulses with 5-ns interval and 71.2-ns pulse. (d) Burst of ten picosecond pulses with 2-ns interval. (e) Burst of ten picosecond pulses with 5-ns interval. (f) Burst of ten picosecond pulses with 10-ns interval. (g) Burst of ten picosecond pulses with 50-ns interval. (h) Single picosecond pulse with 3.3-ms interval. The insets show the corresponding hole exits.
4. Discussion
For nanosecond pulse irradiation (Fig. 5(a)–(b)), heat affected zones (HAZs) were formed with a diameter of ∼100 µm owing to heat accumulation. Meanwhile, picosecond-pulse bursts (Fig. 5(d)–(h)) led to less heat deposition, resulting in smaller HAZs. Exploded molten materials were observed for the 20-ns Gaussian pulse but not for the 100-ns rectangular pulse because the 20-ns Gaussian pulse had a higher intensity (2.15 GW/cm2) than that of the 100-ns rectangular pulse (0.51 GW/cm2). The picosecond-pulse bursts with different pulse intervals produced clear hole exits compared to those produced by nanosecond pulses. Irradiating with multiple laser pulses of different durations and shapes induced multiple ablation as shown in Fig. 5(c). The ablation is due to the combination of drilling by picosecond-pulse bursts and melting by the nanosecond pulse. We note that using different laser parameters simultaneously produces additional effects. In this case, the hole entrance had a smooth surface similar to that observed in the 100-ns rectangular pulse. However, it had a smaller HAZ which was characterized by the picosecond-pulse bursts.
In order to determine the processing speed, the pulse trains in front of and behind the sample were simultaneously measured by two InGaAs photodiodes PD1 and PD2 with a 500-MHz bandwidth as shown in Fig. 1. Figure 6 shows the measured oscilloscope traces of 900 pulses with the 100-ns rectangular pulse shape, and the inset shows the enlarged traces. The penetration time, Δτ, is the time difference of starting points between the input and penetrated pulses as shown in Fig. 6. The processing speed is calculated from the sample thickness divided by the measured penetration time. We assume that the drilled hole has a truncated cone shape, so that the ablated volume can be calculated by the areas of hole entrance and exit, and the sample thickness. Ablation rate, R, which is the material removal rate per pulse is calculated by
where fL is the pulse repetition frequency and V is the ablated volume. The unit is µm3·pls−1.Fig. 6. Representative oscilloscope traces showing the pulse trains measured in front of and behind the 50-µm sample. The results of 900 pulses with the 100-ns rectangular pulse shape are shown, indicating that 19 pulses are required to penetrate the sample. The inset shows the enlarged traces.
We introduce the ablation efficiency, ρ, defined in [17], as the ratio between the minimum energy to transform 1 µm3 from solid to vapor calculated theoretically, called Wth, and the energy required to remove 1 µm3 in the experiment. The ablation efficiency is calculated by the following equation,
Table 1. Processing results performed with different pulse shapes
Laser processes associated with nanosecond pulse had higher ablation efficiencies than those with picosecond-pulse bursts. The shapes of the drilled holes with nanosecond pulse indicated larger taper angles than those with picosecond-pulse bursts. This is because the longer thermal input keeps the material in a molten state for a long period of time. Meanwhile, the results of the picosecond-pulse bursts showed clear hole exits. The taper angle decreased as the pulse interval increased from 2 ns to 10 ns. However, the variability increased as the pulse interval increased. When the pulse interval was set to be 3.3 ms (i.e., 300 Hz repetition frequency), the penetration time fluctuated strongly, resulting in a slow processing speed of less than 1 µm/s. This is understood as being due to the energy accumulation between pulses leading to local heating, which strengthens the coupling between electrons and phonons [15]. The local temperature increase for shorter pulse interval is therefore higher than that for longer pulse interval. The ablation efficiencies on stainless steel for 300–400 femtoseconds pulse increased as the pulse repetition rate increased reported in [17]. Similar tendency was observed for tens of picoseconds pulse duration. The highest ablation rate of 123 µm3·pls−1 was obtained for the 5-ns and 10-ns pulse intervals. The 2-ns pulse interval had an ablation rate of 105 µm3·pls−1, although it had the largest ablated volume. The material removal on the surface was dominant for this case. This can be explained as follows. After the first pulse is irradiated to the surface, the material is melted. Then the second pulse is irradiated to the melted surface, so that the material on the surface is dominantly removed. When the pulse interval is increased to 50 ns, the second pulse is irradiated to the solidified surface. This leads to no exploded molten materials around the drilled hole as shown in Fig. 5(g). The optimal pulse interval depends on thermo-physical properties of materials to be processed.
Combining different laser pulse profiles with large differences in timescales significantly benefits a manufacturing process in which the pre- and post-processes use different types of laser pulses, which could reduce the processing time. Figure 7 shows the SEM images of the drilled holes with and without a post process. Both holes were processed using 300 bursts of ten picosecond pulses with a 2-ns interval and a pulse burst energy of 20 µJ. Additionally, 300 137-ns Gaussian-like pulses with an energy of 40 µJ irradiated the hole as a post process as shown in Fig. 6(b). The burrs caused by the picosecond bursts in Fig. 6(a) were melted and removed by the post laser pulses, resulting in a burr-free surface. The hole exit was also partially melted, reducing the hole size from 8.2 µm to 2.8 µm. Many other pulse patterns can be applied depending on a specific application. The seed pulse generation based on a directly modulated LD technique enables us to combine multiple LDs. Therefore, it is possible to realize a hybrid operation such as a picosecond pulse on top of a nanosecond pedestal.
Fig. 7. Comparison of SEM images with and without a post process. (a) 300 bursts of ten picosecond pulses with 2-ns interval. (b) Condition of (a) followed by 300 137-ns Gaussian-like pulses. The insets show the corresponding hole exits.
5. Conclusion
In summary, we have demonstrated a temporally programmable laser system using a directly modulated LD and Nd:YVO4 amplifiers. The 1-ns temporal resolution of the current driver realized simultaneous generation of picosecond pulses and nanosecond and longer pulses with arbitrary shapes. An experimental demonstration of laser processing of SUS304 with different pulse characteristics has shown that the ablation efficiencies for the nanosecond laser pulses are found to be about a factor of 2 higher than those for the picosecond-pulse bursts. However, the picosecond-pulse bursts exhibited clear hole exits and a relatively small taper angle of less than 2.0°, respectively. Highly flexible laser pulses can potentially be adapted to any type of material. Therefore, this approach will have applications in many areas of industry, including surface modification, spectroscopy, and semiconductor wafer cutting.
Acknowledgement
We thank Mr. Satoru Kobayashi and Mr. Naoki Akiyama for their technical assistance in characterizing the samples.
Disclosures
The authors declare no conflicts of interest.
References
1. W. H. Ko, A. Srinivasa, and P. R. Kumar, “A Multi-Component Automated Laser-Origami System for Cyber-Manufacturing,” IOP Conf. Ser.: Mater. Sci. Eng. 272, 012013 (2017). [CrossRef]
2. V. I. Babushok, F. C. DeLucia Jr, J. L. Gottfried, C. A. Munson, and A. W. Miziolek, “Double pulse laser ablation and plasma: Laser induced breakdown spectroscopy signal enhancement,” Spectrochim. Acta, Part B 61(9), 999–1014 (2006). [CrossRef]
3. A. M. Weiner, J. P. Heritage, and E. M. Kirschner, “High-resolution femtosecond pulse shaping,” J. Opt. Soc. Am. B 5(8), 1563–1572 (1988). [CrossRef]
4. A. M. Weiner, “Femtosecond pulse shaping using spatial light modulators,” Rev. Sci. Instrum. 71(5), 1929–1960 (2000). [CrossRef]
5. P. Xi, Y. Andegeko, D. Pestov, V. V. Lozovoy, and M. M. Dantus, “Two-photon imaging using adaptive phase compensated ultrashort laser pulses,” J. Biomed. Opt. 14(1), 014002 (2009). [CrossRef]
6. K. A. Mumtaz and N. Hopkinson, “Selective laser melting of thin wall parts using pulse shaping,” J. Mater. Process. Technol. 210(2), 279–287 (2010). [CrossRef]
7. S. Banerjee, P. Mason, K. Ertel, P. Jonathan Phillips, M. De Vido, O. Chekhlov, M. Divoky, J. Pilar, J. Smith, T. Butcher, A. Lintern, S. Tomlinson, W. Shaikh, C. Hooker, A. Lucianetti, C. Hernandez-Gomez, T. Mocek, C. Edwards, and J. Collier, “100J-level nanosecond pulsed diode pumped solid state laser,” Opt. Lett. 41(9), 2089–2092 (2016). [CrossRef]
8. C. E. Rogers and P. L. Gould, “Nanosecond pulse shaping at 780 nm with fiber-based electro-optical modulators and a double-pass tapered amplifier,” Opt. Express 24(3), 2596–2606 (2016). [CrossRef]
9. M. Nie, Q. Liu, E. Ji, X. Cao, X. Fu, and M. Gong, “Active pulse shaping for end-pumped Nd:YVO4 amplifier with high gain,” Opt. Lett. 42(6), 1051–1054 (2017). [CrossRef]
10. R. A. Meijer, A. S. Stodolna, K. S. E. Eikema, and S. Witte, “High-energy Nd:YAG laser system with arbitrary sub-nanosecond pulse shaping capability,” Opt. Lett. 42(14), 2758–2761 (2017). [CrossRef]
11. K. T. Vu, A. Malinowski, D. J. Richardson, F. Ghiringhelli, L. M. B. Hickey, and M. N. Zervas, “Adaptive pulse shape control in a diode-seeded nanosecond fiber MOPA system,” Opt. Express 14(23), 10996–11001 (2006). [CrossRef]
12. H. Ito, H. Yokoyama, S. Murata, and H. Inaba, “Picosecond optical pulse generation from an R.F. modulated AlGaAs double heterostructure diode laser,” Electron. Lett. 15(23), 738–740 (1979). [CrossRef]
13. H. Liu, C. Gao, J. Tao, W. Zhao, and Y. Wang, “Compact tunable high power picosecond source based on Yb-doped fiber amplification of gain switch laser diode,” Opt. Express 16(11), 7888–7893 (2008). [CrossRef]
14. T. Chen, H. Liu, W. Kong, and R. Shu, “Burst-mode-operated, sub-nanosecond fiber MOPA system incorporating direct seed-packet shaping,” Opt. Express 24(18), 20963–20972 (2016). [CrossRef]
15. W. Hu, Y. C. Shin, and G. King, “Modeling of multi-burst mode pico-second laser ablation for improved material removal rate,” Appl. Phys. A: Mater. Sci. Process. 98(2), 407–415 (2010). [CrossRef]
16. Y. Yokoyama, K. Takada, T. Kageyama, S. Tanaka, H. Kondo, S. Kanbe, Y. Maeda, R. Mochida, K. Nishi, T. Yamamoto, K. Takemasa, M. Sugawara, and Y. Arakawa, “2014-nm DFB laser diode modules applicable to seeder for pulse-on-demand fiber laser systems,” Opt. Fiber Technol. 20(6), 714–724 (2014). [CrossRef]
17. J. Lopez, G. Mincuzzi, R. Devillard, Y. Zaouter, C. Hönninger, E. Mottay, and R. Kling, “Ablation efficiency of high average power ultrafast laser,” J. Laser Appl. 27(S2), S28008 (2015). [CrossRef]
